Machine Learning Algorithms for atrioventricular conduction defects prediction using ECG: A comparative study

Abstract

An electrocardiogram is a propitious tool for the diagnosis of myriad cardiac diseases such as atrioventricular blocks. The abnormal activity of the heart can be detected using leads which record electric signals generated by the heart. A preliminary study effectuated for single-lead electrocardiograms exhibited the superiority of machine learning models. Therefore, we performed a comparative study using ECG-derived Data from the KURIAS-ECG database to analyze which machine learning algorithm or neural network model can detect atrioventricular conduction defects and categorize them with better accuracy. To this effect, we have made utilization of three models: Gaussian Naive Bayes Function, Random Forest Classifier, NeuralNetwork with One-Hot Encoding. This study conducted by the authors will thus aid in the selection of the most suitable model for the detection and categorization of these defects.